Paper 13305-86
Paper 13305-86
Deep reinforcement learning for automatic focus and axial motion correction for OCT B-scans
On demand | Presented live 26 January 2025 
Abstract
Optical coherence tomography (OCT) retinal imaging often faces challenges from subject movement and focus accommodation, disrupting consistent imaging of targeted layers. We present a method utilizing OCT B-scan images and a deep reinforcement learning (DRL) algorithm to correct focus changes and axial shifts. Trained in a simulated environment and refined with transfer learning, the DRL agent provides consistent retinal layer focus and stable imaging.
Presenter
Guozheng Xu
Univ. College London (United Kingdom)
Guozheng Xu is currently pursuing his PhD at the EPSRC Centre for Doctoral Training in Intelligent, Integrated Imaging in Healthcare (i4health) at University College London. He holds a Bachelor's degree in Optoelectronic Engineering from Huazhong University of Science and Technology and a Master of Research degree from the University of Cambridge.
Guozheng's research is dedicated to advancing novel biomedical imaging technologies, focusing on ophthalmology applications. His work includes developing and enhancing scanning laser ophthalmoscopes, optical coherence tomography, and adaptive optics systems.
Also in the session